Revisit hydrological modeling in ungauged catchments comparing regionalization, satellite observations, and machine learning approaches

Abstract

Understanding hydrological processes is achieved using modeling approaches due to the extensive and complex interactions between various environmental elements. Hydrological modeling is based on empirical equations that require parameter calibration and model validation to improve performance and evaluate results. This process requires the implementation of absent or lacking data in many ungauged catchments. Therefore, Hydrological Modeling in Ungauged Catchments (HMUC) is an important research area in hydrology. Many researchers tried to develop appropriate technology for this purpose. This review article describes regionalization, satellite observation and machine learning based technologies used for this purpose and presents relevant issues. Key studies worldwide using regionalization, satellite observations and machine learning approaches to develop HMUC have been reviewed here. This study promotes research on HMUC by describing the performances of these methods in different climatic, and geographic conditions. It identifies potential application limitations to guide the framing of future requirements and opportunities for HMUC.

Keywords

• Hydrological model
• Ungauged catchment
• Regionalization
• Satellite-based discharge
• Long short-term memory network